Age | Commit message (Collapse) | Author |
|
X-Git-Url: http://git.openmoko.org/?p=kernel.git;a=commitdiff_plain;h=f8fbb07854e3aff64dce9fe6ef6a8dc0e0f762b5
spi_supoport_non_blocking_sync_transfers.patch
A mew option is added to spi_bitbang_info to specify if the
transfers off it will be non blocking.
A new function - spi_non_blocking_transfer is added to the SPI core.
Signed-off-by: Balaji Rao <balajirrao@openmoko.org>
|
|
Hi Andy,
Here's the fix for the qvga-normal bug we observed the other day
here - when the screen is powered up again after a blanking, in
qvga-normal mode, we see a WSOD(kind of!)
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
This adds in the TZIC support consts and also the VM mapping
for the TZIC units, and adds TZIC init into mach-om-gta03.c
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Clean out more switch() on version that has no point given the
actual features that (don't) differ on supported GTA02 versions.
We take care about forcing GPS regulator off on neo1973-pm-gps probe,
but we don't take care about killing the UART GPIO so they don't
drive into the powered-down subsystem. We took care of doing that when
we turned off the thing already, but we cannot just call the OFF code
from the probe due to problems with regulator state init.
Reported-by: Rask Ingemann Lambersten <rask@sygehus.dk>
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Drop the i2c0 bus max-frequency to 150kHz to avoid
problems with boards where this is too fast.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
--- linux.agit.orig/arch/arm/plat-s3c/dev-i2c0.c 2009-02-20 23:04:34.000000000 +0000
+++ linux.agit/arch/arm/plat-s3c/dev-i2c0.c 2009-02-20 23:04:37.000000000 +0000
@@ -51,7 +51,7 @@ static struct s3c2410_platform_i2c defau
.flags = 0,
.slave_addr = 0x10,
.bus_freq = 100*1000,
- .max_freq = 400*1000,
+ .max_freq = 150*1000,
.sda_delay = S3C2410_IICLC_SDA_DELAY5 | S3C2410_IICLC_FILTER_ON,
};
--
Ben (ben@fluff.org, http://www.fluff.org/)
'a smiley only costs 4 bytes'
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Because i found the level of the "MODEM_RST" will return to "L" when it set to "Input".
GSM module sometimes can not turn on especially after emove the C9102.
Signed-off-by: DKAY_CHEN <dkay_chen@openmoko.com>
|
|
Ensure the DMA channels have the controller clock
enabled before they are used.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
|
|
Add the DMA channels as system devices to provide
suspend/resume and a presence in sysfs.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
|
|
Add the clock sources for the DMA and SDMA controllers
and ensure that they are shut down at startup. The DMA
engines are dma0 and dma1, whilst the SDMA are called
dma2 and dma3.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
|
|
Fix the definitions of HCLK_GATE bits, which
had SDMA0/1 named SDMA1/2 and the definitions
of bits 26..29 in the wrong order.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
hes-tracking-s3c64xx-dma-support-1235439162-1235439227
pending-tracking-hist top was MERGE-via-stable-tracking-MERGE-via-mokopatches-tracking-s3c64xx-dma-support-1235439162-1235439227 / 3d6a1b21cf5fbdb6250d781b0a4900a7a0768aa1 ... parent commitmessage:
From: merge <null@invalid>
MERGE-via-stable-tracking-hist-MERGE-via-mokopatches-tracking-s3c64xx-dma-support-1235439162
stable-tracking-hist top was MERGE-via-mokopatches-tracking-s3c64xx-dma-support-1235439162 / 893e864e65adffc9eb085ed4f8b552a31dcec840 ... parent commitmessage:
From: merge <null@invalid>
MERGE-via-mokopatches-tracking-hist-s3c64xx-dma-support
mokopatches-tracking-hist top was s3c64xx-dma-support / 2515f9a1d53d19b1e61d639875aedcbe7929666e ... parent commitmessage:
From: Ben Dooks <ben@simtec.co.uk>
S3C64XX: DMA support
Add support for the DMA blocks in the S3C64XX series
of CPUS, which are based on the ARM PL080 PrimeCell
system.
Unfortunately, these DMA controllers diverge from the
PL080 design by adding another DMA controller register
and configuration for OneNAND.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
It's always a good idea to first define the data structures you use. ;-)
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
For GTA02 we can remove GTA02_HDQ (not used anywhere) and
S3C2410_CLOCK (already selected by CPU_S3C2442).
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
This reverts commit fe8b7ff1a79648146501f37d467fcb2f2bea0164.
I undo the reordering of the options here, as it hides what
really happend.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
For GTA02 we can remove GTA02_HDQ (not used anywhere) and
S3C2410_CLOCK (already selected by CPU_S3C2442). Additionally I reordered
the options a bit.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
If the PCF50633 ADC is disabled, but the charger is enabled we can't determine
the charger type using the pulldown. The safest thing we can do is to fallback
to 100mA that should be supplied according to the USB specs.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
If the vibrator is disabled we don't need to register a platform device.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
We select the PCF50633 core part, which depends on I2C, in our Kconfig. However,
make oldconfig does not automatically resolve this "depends on I2C" dependency.
We should select I2C for GTA02 as it's needed anyway.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
The GSM interface of all MACH_NEO1973s require the Samsung serial console,
so we should automatically select it.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
Move the battery section within the HDQ_GPIO_BITBANG ifdef because it uses
functions only available if HDQ bitbanging is enabled.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
If we disable CONFIG_CHARGER_PCF50633 the build fails with a bunch of errors.
This patch defines placeholders for the functions that require the charger. All
users of the function pointers care about the case when they are NULL, so this
won't break anything. Furthermore a small restructuring of the mach file was
necessary to not sprinkle ifdefs all over.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
Remove the old config option and use the correct new one instead.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
We have this in the S3C2442 machine section now.
Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>
|
|
Hi,
since I am using packaged cross-compiler from debian I need to change
export CROSS_COMPILE=/usr/local/openmoko/arm/bin/arm-angstrom-linux-gnueabi-
in ./build to
export CROSS_COMPILE=arm-angstrom-linux-gnueabi-
every time there are changes to ./build and also instruct others to do
this when they want to test their kernels with the "normal
cross-compiler" and not the one from openmoko. How about making it
possible to override CROSS_COMPILE with for example the following
change?
|
|
This patch cleans some cruft from neo1973-pm-gsm.c and makes
some changes to the power code for gsm.
First, on probe it enforces GSM is OFF.
Second, when GSM is OFF, it changes the RTS and TXD pins on
UART 0 to be inputs, instead of driving into a dead subsystem
and costing us ~70mA @ 5V constant dissipation (these 3.3V pins
were found at 1.5V due to that).
Third it asserts the GPIO peripheral function when ON, to keep
the same flow for the new interrupt generation workaround code.
Since several people worked on this area before, and maybe the
assertion of GSM OFF is different (it's OFF from bootloader though)
or maybe something else gets broken, I send the patch for comment
before applying it.
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Added support for HDQ to GTA03.
Signed-off-by: Balaji Rao <balajirrao@openmoko.org>
|
|
Signed-off-by: Balaji Rao <balajirrao@openmoko.org>
|
|
Introduce a new generic HDQ gpio based bitbang driver
Signed-off-by: Balaji Rao <balajirrao@openmoko.org>
|
|
Since the pwm code in mach-2410 can be reused for 6410, move it to plat-s3c.
Signed-off-by: Balaji Rao <balajirrao@openmoko.org>
|
|
The code for creating FIQ handlers in C is moved to
arch/arm/kerne/fiq.c
Signed-off-by: Balaji Rao <balajirrao@openmoko.org>
|
|
This implements a proper scheme of controlling powering on of the Calypso
GSM modem which should be activated with a "push-button" connected to
MODEM_ON line.
Also a workaround for firmware bug of moko10 and earlier implemented to
force calypso to pull IRQ line down after powering on.
The "reset" sysfs node is left for compatibility with userspace (without it
gsm0710muxd breaks) but it does nothing (NC on hardware level on GTA02
anyway).
I tested flashing the GSM firmware (on GTA02) with this patch (using only
fluid and power_on node) and can confirm that it's working as reliably as
with accessing gpio directly per instructions.
The credit goes to Joerg Reisenweber for discussions and clarifications on
how this should be done properly.
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
|
|
Pardon me if I am too verbose but this is my first opportunity to send a patch.
I was wishing to try andy-tracking kernels on GTA01 long ago but It
was FOSDEM and recent threads on the lists asking for GTA01 that made
me do it.
When I participated on the thread
http://n2.nabble.com/empiezo-a-amargarme-con-los-kernels-necesito-consejo-td2347206.html
in the Spanish mailing list asking what kernel to use with which
distribution, I responded that only FSO based supported 2.6.28 and
upwards. I could not resist to brag that I tested 2.6.29 on GTA01 but
only with kexec ( it is possible that my wording was more ambiguous on
the thread).
Nelson Asked if I had problems booting it with u-boot. I tried and was
successful but this time I played a Little with the interface and vala
term and noticed that the on screen keyboard was changing the key
pressed.
Using http://svn.openmoko.org/developers/tick/touchtest/touch_test.py
shown that most of the Up and sometimes the previous Move to where
landing on ([290..310],575).
Revising the code shown that the some of the filters and the option to
disable them were missing on mach-gta01.c.
I have tested the kernel with the patch applied with the filters
disabled and also with them enabled. When the filters are active I
still receive some misplaced up events but much less (the raster
terminal keyboard is usable).
Signed-off-by:Jose Luis Perez Diez <perezdiez@gmail.com>
|
|
This is a known problem and it is starting to become a FAQ.
People get an error message when they try to build a kernel
and they are not in a valid branch of their local repository.
When it happens git-branch will print something like:
* (no branch)
andy-tracking
balaji-tracking
With this patch we make sure we are in a valid branch when
building the kernel using the build script in order to avoid
OM developers some trouble. How? We print a nice message if
we are not in a valid branch and then stop.
Please feel free to edit the error message if you think it is
not clear enough.
Thanks to Werner Almesberger for his advice on shell scripting.
Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>
|
|
This patch moves the bulk transfer action outside of
interrupt context, along with the STOP transmission action
for multiblock transfers.
It's prompted by
https://docs.openmoko.org/trac/ticket/2180
But it can impact throughput to SD card, so it's for testing
currently.
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
Signed-off-by: Andy Green <andy@openmoko.com>
|
|
This strips out the v1 and v2 GTA02 revision specific defines
(we do not support these internal non-shipped revisions) and
fixes a bug in the glamo reset resource found from that.
Signed-off-by: Andy Green <andy@openmoko.com>
|